Large language models (LLMs) represent the currently most relevant incarnation of artificial intelligence with respect to the future fate of democratic governance. Considering their potential, this paper seeks to answer a pressing question: Could LLMs outperform humans as expert advisors to democratic assemblies? While bearing the promise of enhanced expertise availability and accessibility, they also present challenges of hallucinations, misalignment, or value imposition. Weighing LLMs’ benefits and drawbacks compared to their human counterparts, I argue for their careful integration to augment (...) democracy’s ability to address complex policy issues. The paper posits that time-tested democratic procedures like deliberation and aggregation by voting provide safeguards effective against both human and machine advisor imperfections. Additional protective measures include custom LLM training for the advisory role, boosting representatives’ competencies in query formulation, or implementation of adversarial proceedings in which LLM advisors could debate each other and provide dissenting opinions. These could further mitigate the risks that LLMs present in advisory roles and empower human decision-makers toward increased autonomy and quality of their collective choices. My conceptual exploration offers a roadmap for the co-evolution of AI and democratic institutions, setting the stage for an empirical research agenda to finetune the implementation specifics. (shrink)

ABSTRACT The “diversity trumps ability” model is often interpreted as a mechanism supporting epistemic democracy. However, as a variety of empirical and mathematical studies have shown, if we attempt to test the realism of the model, it turns out that it points as much toward epistocracy as democracy. This might appear to leave epistocracy with an advantage, since its rationale is not usually thought to rely on the DTA but on the obvious relevance of expertise to making complex decisions. Yet (...) if we apply the same test to epistocracy that we should apply to epistemic democracy—the test of realism—we find that it, too, is unsustainable. This suggests that epistemic democracy and epistocracy alike are indefensible on the basis of the abstract assumptions about diversity and expertise on which the DTA is predicated. (shrink)

Supposedly, stubbornness on the part of scientists—an unwillingness to change one’s position on a scientific issue even in the face of countervailing evidence—helps efficiently divide scientific labor. Maintaining disagreement is important because it keeps scientists pursuing a diversity of leads rather than all working on the most promising, and stubbornness helps preserve this disagreement. Planck’s observation that “Science progresses one funeral at a time” might therefore be an insight into epistemically beneficial stubbornness on the part of researchers. In conversation with (...) extant formal models, recent empirical research, and a novel agent-based model of my own I explore whether the epistemic goods which stubbornness can secure—disagreement and diversity—are attainable through less-costly methods. I make the case that they are, at least in part, and also use my modeling results to show that if stubbornness is scientifically valuable, it still involves a willingness to change one’s mind. (shrink)

This article contributes to the revision of the procedure of robustness analysis of mathematical models in epistemic democracy using the systematic review method. It identifies the drawbacks of robustness analysis in epistemic democracy in terms of sample universality and inference from samples with the same results. To exemplify the effectiveness of systematic review, this article conducted a pilot review of diversity trumps ability theorem models, which are mathematical models of deliberation often cited by epistemic democrats. A review of nine models (...) extracted from 352 papers exemplifies the effectiveness of robustness analysis supplemented by systematic review in epistemic democracy. (shrink)

According to the diversity-beats-ability theorem, groups of diverse problem solvers can outperform groups of high-ability problem solvers. We argue that the model introduced by Lu Hong and Scott Page is inadequate for exploring the trade-off between diversity and ability. This is because the model employs an impoverished implementation of the problem-solving task. We present a new version of the model that captures the role of ‘ability’ in a meaningful way, and we use it to explore the trade-offs between diversity and (...) ability in scientific problem solving. (shrink)

There have been recent calls to expand contemporary analytic philosophy of religion beyond the oft implicitly assumed Christian tradition. Instead of exploring moral reasons to expand the discipline, I argue that there are strong epistemic reasons to favour diversifying the philosophy of religion. Increasing diversity is likely to increase disagreement, and there are epistemic benefits to be gained from the existence of disagreement. I argue that such considerations quite clearly apply to the philosophy of religion, and as such that there (...) are epistemic reasons to diversify the field. I conclude by offering a number of practical steps we can take towards achieving this end which are relatively easy to implement. (shrink)

Epistemic paternalism occurs when a decision is made for an agent which helps them arrive at the truth, though they didn’t consent to that decision (and sometimes weren’t even aware of it). Common defenses of epistemic paternalism claim that it can help promote positive veritistic results. In other words, epistemic paternalism is often good for inquiry. I argue that there is often a better alternative available to epistemic paternalism in the form of what I call Open Group Inquiry. I then (...) examine how Open Group Inquiry can be applied to cases of religious inquiry, while noting that epistemic paternalism is impermissible in cases of general religious inquiry. I argue that in the case of religious inquiry, there are serious questions about what constitutes evidence along with how to evaluate it. Rather than posing a particular worry for Open Group Inquiry, I suggest these questions pose a problem for religious inquiry in general. I conclude that while it very much matters how concepts like religious knowledge, religious faith, scepticism, etc., are defined, these considerations may well pave the way for a novel argument for religious scepticism. (shrink)

Epistemic justifications for democracy have been offered in terms of two different aspects of decision-making: voting and deliberation, or ‘votes’ and ‘talk.’ The Condorcet Jury Theorem is appealed to as a justification in terms votes, and the Hong-Page “Diversity Trumps Ability” result is appealed to as a justification in terms of deliberation. Both of these, however, are most plausibly construed as models of direct democracy, with full and direct participation across the population. In this paper, we explore how these results (...) hold up if we vary the model so as to reflect the more familiar democratic structure of a representative hierarchy. We first recount extant analytic work that shows that representation inevitably weakens the voting results of the Condorcet Jury Theorem, but we question the ability of that result to shine light on real representative systems. We then show that, when we move from votes to talk, as modeled in Hong-Page, representation holds its own and even has a slight edge. (shrink)

The Hong and Page ‘diversity trumps ability’ result has been used to argue for the more general claim that a diverse set of agents is epistemically superior to a comparable group of experts. Here we extend Hong and Page’s model to landscapes of different degrees of randomness and demonstrate the sensitivity of the ‘diversity trumps ability’ result. This analysis offers a more nuanced picture of how diversity, ability, and expertise may relate. Although models of this sort can indeed be suggestive (...) for diversity policies, we advise against interpreting such results overly broadly. (shrink)

Peer review is often taken to be the main form of quality control on academic research. Usually journals carry this out. However, parts of maths and physics appear to have a parallel, crowd-sourced model of peer review, where papers are posted on the arXiv to be publicly discussed. In this paper we argue that crowd-sourced peer review is likely to do better than journal-solicited peer review at sorting papers by quality. Our argument rests on two key claims. First, crowd-sourced peer (...) review will lead on average to more reviewers per paper than journal-solicited peer review. Second, due to the wisdom of the crowds, more reviewers will tend to make better judgments than fewer. We make the second claim precise by looking at the Condorcet Jury Theorem as well as two related jury theorems developed specifically to apply to peer review. (shrink)

Two types of formal models - landscape search tasks and two-armed bandit models - are often used to study the effects that various social factors have on epistemic performance. I argue that they can be understood within a single framework. In this unified framework, I develop a model that may be used to understand the effects of functional and demographic diversity and their interaction. Using the unified model, I find that the benefit of demographic diversity is most pronounced in a (...) functionally homogeneous group, and decreases with the increase of functional diversity. (shrink)

According to the diversity-beats-ability theorem, groups of diverse problem solvers can outperform groups of high-ability problem solvers. We argue that the model introduced by Lu Hong and Scott Page is inadequate for exploring the trade-off between diversity and ability. This is because the model employs an impoverished implementation of the problem-solving task. We present a new version of the model which captures the role of ‘ability’ in a meaningful way, and use it to explore the trade-offs between diversity and ability (...) in scientific problem solving. (shrink)